Should an AI answering your calls say it's an AI?

People ask. Not often, but they ask. "Am I talking to a real person?" The tempting answer for a voice product is to dodge. Ours is to say yes to the question and no to the dodge: Eva tells the caller she is a virtual assistant. This post is about why, and about what the law requires so far.

California's bot law

California's Bolstering Online Transparency Act, SB 1001, has been in force since July 1, 2019. It makes it unlawful to use a bot to communicate with a person in California with the intent to mislead them about its artificial identity in order to encourage a purchase or influence a vote. The way out is a disclosure that is clear, conspicuous and reasonably designed to inform the person they are dealing with a bot. The law is written for online communication and is enforced by the Attorney General; there is no private right to sue under it.

The FCC and AI voices

On February 8, 2024 the Federal Communications Commission issued a declaratory ruling that AI generated voices count as "artificial" under the Telephone Consumer Protection Act. In plain terms, an outbound call that uses a synthetic voice is treated like a prerecorded robocall: it needs the recipient's prior express consent, the caller has to identify itself, and there has to be a way to opt out. The ruling came a month after a cloned voice of the President was used in robocalls before the New Hampshire primary.

That ruling is about outbound calls to people in the United States. It does not tell an inbound assistant what to say when someone rings a plumber. But it is the clearest signal yet of where regulators are going: people have a right to know when they are hearing a machine.

Canada

Canada has no bot disclosure statute of its own as of this writing. Businesses are still bound by the privacy law, PIPEDA, and by the CRTC's rules on unsolicited calls, which require callers to identify themselves and restrict calls to people who have not done business with you. None of that mentions synthetic voices. It also does not need to, for the same reason the California law exists: pretending to be a person to win a sale was already a bad idea.

What we do

• Eva answers as Eva. Asked directly whether she is a person or an AI, she says she is a virtual assistant for the business.
• Max calls back people who asked the business for a quote. He says who he is calling for and why in the first sentence, and he stops if they ask him to.
• Every call is recorded, and the caller is told so. The consent stays with the recording.

Honesty here is not a compliance cost. A caller who finds out later that the friendly receptionist was a machine has been lied to by your business. Nobody books the reroof with a business that lied to them on the first call.